Todd Kolber, co-managing director of Logistics Reply, notes that agentic artificial intelligence is improving the operations of warehouse management systems (WMS). According to Kolber's perspective, current WMS architecture is insufficient. Planners look months or even years ahead, operations are tasked with meeting these forecasts, and management must serve as the "traffic cop" keeping everything running. From a supply chain perspective, WMS, along with WCS (Warehouse Control System) and WES (Warehouse Execution System), form three main layers of the warehouse stack. Manhattan Associates, Blue Yonder, SAP EWM, Oracle WMS Cloud, Körber, Softeon, and Logistics Reply are the market's major players, and all are moving toward agentic AI integration.
Kolber states: "The problem that a warehouse management system has is that it can't see the whole." Kolber adds: "It's answering a need for a given task and fulfilling orders, but it's not really providing what is the total optimal outcome for a process or facility. There's still a lot to be desired today." Either management or a solution must be far more aware at the "umbrella level" about all resources, functions, and technologies across the entire facility. Kolber emphasizes: "The key word is orchestrate – orchestrate what happens in that facility." From a supply chain perspective, warehouse orchestration encompasses the synchronized coordination of AGV (Automated Guided Vehicle), AMR (Autonomous Mobile Robot), AS/RS (Automated Storage and Retrieval System), conveyor, sorter, and human labor. Automation providers such as Locus Robotics, 6 River Systems, Geek+, AutoStore, and Symbotic offer an API-first approach for WMS integration points.
So much data and activity happens in real time that it becomes impossible for any individual or team to make the right decision about what action should be taken. Kolber states: "It's going to be critical that AI and agentic AI systems play a major role in bringing all that data together, analyzing it, looking at historical trends and patterns, then making decisions based on all that information in real time and providing the optimal answer." More importantly, when unexpected things happen—when a large order comes in, when a machine begins performing at historically lower speeds, or when an area of the warehouse goes down—people cannot react fast enough to recover. From a supply chain perspective, real-time event processing runs on streaming architecture based on Apache Kafka, Apache Flink, and Confluent. Digital twin technology, via platforms such as Siemens Tecnomatix Plant Simulation, FlexSim, and AnyLogic, simulates what-if scenarios and provides the foundation for agentic AI action decisions.
Kolber states: "AI can analyze all that data in real time, solve things in such a way that you can still meet the goals of that day, and do it at optimal cost." From a supply chain perspective, cost-to-serve optimization requires real-time recalculation of the balance between labor cost, equipment utilization, energy consumption, and SLA (Service Level Agreement) penalty costs. For 3PL (Third-Party Logistics) operators, this approach enables simultaneous optimization of different customer requirements in multi-tenant facilities. Logistics Reply, as part of the Reply Group and headquartered in Italy, serves a global customer base across EMEA, Americas, and APAC. In conclusion, Kolber's vision signals that the WMS market will evolve over the next 3–5 years from order management alone to integrated real-time facility orchestration.
Key Takeaways:
1. Kolber notes that current WMS architecture is insufficient.
2. WMS cannot see the whole picture; it operates only on a task basis.
3. The concept of "orchestrate" is key to coordinating all resources and technologies.
4. Real-time data exceeds human decision-making capacity; AI and agentic AI play a critical role.
5. AI enables achievement of daily goals at optimal cost.
